+ All Categories
Home > Health & Medicine > Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized...

Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized...

Date post: 06-May-2015
Category:
Upload: gunther-eysenbach
View: 2,117 times
Download: 0 times
Share this document with a friend
19
Bamidis, P. et al.: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice • This slideshow, presented at Medicine 2.0’08, Sept 4/5 th , 2008, in Toronto, was uploaded on behalf of the presenter by the Medicine 2.0 team • Do not miss the next Medicine 2.0 congress on 17/18th Sept 2009 (www.medicine20congress.com ) • Order Audio Recordings (mp3) of Medicine 2.0’08 presentations at http://www.medicine20congress.com/mp3.php
Transcript
Page 1: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Bamidis, P. et al.:Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice

• This slideshow, presented at Medicine 2.0’08, Sept 4/5th, 2008, in Toronto, was uploaded on behalf of the presenter by the Medicine 2.0 team

• Do not miss the next Medicine 2.0 congress on 17/18th Sept 2009(www.medicine20congress.com)

• Order Audio Recordings (mp3) of Medicine 2.0’08 presentations at http://www.medicine20congress.com/mp3.php

Page 2: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Charalampos Bratsas, Panagiotis Bamidis *, Evangelos Kaimakamis, Nicos Maglaveras

Lab of Medical Informatics, Medical School

Aristotle University of Thessaloniki

Usage of Semantic Web Usage of Semantic Web Technologies (Web-3.0) Aiming to Technologies (Web-3.0) Aiming to

Facilitate the Facilitate the UtilisationUtilisation of of Computerized Algorithmic Computerized Algorithmic

Medicine in Clinical PracticeMedicine in Clinical Practice

Page 3: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

OutlineOutline

Definition of Medical Computational Problems and the benefits of use algorithms in Medicine

Why algorithmic medicine doesn't used? What is the main problem?

Scope – Solutions Ontologies as a structure framework of MCPs Methods and Web-System architecture (KnowBaSICS-

M) Experimental evaluation and test case Future research

3C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 4: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

MedicalMedical Computational ProblemsComputational Problems – – Computerized Algorithmic SolutionsComputerized Algorithmic Solutions

Medical Computational Problems MCPs: Medical problems, the solution of which deals with mathematical or statistical models, signal or image processing and estimation of corresponding parameters.

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Screen a patient for sleep apnea using the multivariable apnea risk index

Medical ProblemClinical Algorithm

Implementation

Medical Computational Problem

u2

x3

x4

u1

x2

x1

f(x1...xn)

A11

A11

A11

A11

A11

A11

Maislin et.alscore for predicting sleep apnea in patients

To define MCPs and their solutions diffe

rent domains

of knowledge are required

Collaboration of different kind of scientists.

Page 5: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Conclusions of MIE 2006 WorkshopConclusions of MIE 2006 Workshop

There are tens of thousands of algorithms. They are not widely incorporated into routine care. We believe that healthcare would be better if they

were. Ontology support for Algorithmic Medicine

John R Svirbely, Jan Vejvalka, M Sriram Iyengar, Charalampos Bratsas, Evangelos Kaimakamis, Nicos Maglaveras.

Technological guidelines for integrating medical algorithms into healthcare systems

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 6: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Conclusions of MIE 2006 WorkshopConclusions of MIE 2006 Workshop

Why aren’t algorithms used? ◦ I don’t have the time.◦ I didn’t know there was one.◦ I don’t remember what it is.◦ I don’t have a software.◦ I don’t have the data I need.◦ I don’t know how to use it.

John R Svirbely, Jan Vejvalka, M Sriram Iyengar, Charalampos Bratsas, Evangelos Kaimakamis,Nicos Maglaveras.

Technological guidelines for integrating medical algorithms into healthcare systems

Page 7: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Main reason -Solution Main reason -Solution

Doctors, Doctors, Mathematicians, Mathematicians, PhysicsPhysics , etc, etc

InformaticInformaticss

Structure Framewor

k to describe MCPs Ontologie

s

Structure and EducationC Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 8: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Scope - SolutionsScope - Solutions

Develop the semantic framework (MCP Ontology]) enclosing the required knowledge based on which the medical problem - algorithm - implementation are semantically described.

Develop knowledge retrieval methods, through ontological questions and the utilization of information retrieval methods inside the MCP Ontology.

Develop dynamic semantic composition of a sequence of algorithms managing a certain medical case

Scope:The initial development of semantic descriptions of Medical Computational Problems (MCPs) and the management of resulting knowledge.

Page 9: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

MCP OntologyMCP Ontology

The MCP Ontology is an OWL ontology model that manages MCPs and their solutions by means of organizing and visualizing their existing knowledge.

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 10: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

MCP Ontology Model MCP Ontology Model

Ontologies :◦ Medical

Problem Ontology

◦ Medical Algorithm Ontology

◦ Implementation Ontology

◦ Users Ontology

Reuses or/and Adaptations:•BibTex Ontology to semantically describe the MCPs References (http://www.cs.toronto.edu/semanticweb/maponto/ontologies/BibTex.owl )

•UMLS Ontology to semantically describe the medical concepts. (Unified Medical Language System) (http://umlsks.nlm.nih.gov/kss)

•ConOnto Ontology to semantically describe the software and hardware of implemented algorithm (http://www.site.uottawa.ca/~mkhedr/Ontologies/ )

•Global Medical Device Nomenclature to semantically describe the medical devices (http://www.gmdnagency.com/) 10

Page 11: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Adaptation of the classical Vector Space

Model (VSM) in MCP Ontology based on which1. The MCP weighted vectors are created by the

implementation of the weights of the UMLS terms acting as the problem indexing terms in the MCP Ontology

tf factor: based on the frequency of occurrence of the instances of a keyword (UMLS concept) into MCPs natural language description

idf factory: based on frequency of occurrence of the instances of a keyword (UMLS concept) into MCP Ontology.

2. The similarity between MCP semantic descriptions and the user questions is calculated.

Cosine Similarity

MCP Ontology – MCP Ontology – Efficient SearchEfficient Search

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 12: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

MCP Ontology - Managing a MCP Ontology - Managing a certain medical casecertain medical case

◦ Dynamic semantic composition of a sequence of algorithms1.Using semantic rules, the links between different

algorithms are created and used in the construction of a Finite State Machine (FSM) of algorithms. 1st Set of Rules: Define the Possible Prerequisites

Algorithms of an algorithmic solution. (Input/Output Variables)

2nd Set of Rules: Define the Possible Related Algorithms of an algorithmic solution. (Output/Output Variables)

2.Description of a certain medical case via the MCP Ontology by a user constitutes the language of that case which is recognised by a FSM of algorithms with the final algorithm managing the case as the initial state and the algorithm of initiation by the user as the final state. Set of Rules: Define the Available Algorithmic Solutions for

a specific medical case (Pre-conditions are met)

Page 13: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Knowledge Insertion and Redefine

Module

MCP KB

Semantic MCP Repository

MCP Ontology Model

Encodes the MCP semantic model andprovides knowledge

acquisition

Query Formulator

Query Processor

Result Set Retrieval

Query Engine

Similarity Calulator

Vectors Constructor

Ontology VSM

Create Vectors fromInstances of the extractedTerms and calculates similarity metrics

Medical Concepts Extractor

Medical Terms Annotator

UMLS-KB

UMLS-based medical concepts extraction on the

user define query

Medical Case Management Planning

FSM Constructor

Algorithmic Solution Sequence Designer

Creates a FSM of available algorithms for specific

medical case Retrieve the algorithmic sequence that manages

the specific case

Performs Ontology based (SPARQL) knowledge retrievalon MCPS

User Interface

13

KnowBaSICS-M Modular Architecture KnowBaSICS-M Modular Architecture

Page 14: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

KnowBaSICS-M Technical Architecture KnowBaSICS-M Technical Architecture DiagramDiagram

Code development was based on open-source development platforms and tools: (Protégé, Java, Jena, eclipse, Millstone)

The system consists of:◦ MCP Management Server◦ 2 Clients

Java Standalone Web Client

14C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 15: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Experimental evaluationExperimental evaluation - - GoalsGoals

To evaluate KnowBaSICS-M either for knowledge insertion or for knowledge retrieval in order to assess its usability.

To calculate the precision and recall features. To evaluate KnowBaSICS-M to manage

specific cases by dynamically semantic composite algorithmic sequences

15C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 16: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Evaluation ProcessEvaluation Process

Process Layer

Result Layer

DecisionLayer

Criteria Layer

Phase 1 Phase 2 Phase 3 Phase 4

1. Web Sites with MCPs 2.Well, no structure – no

semantic description3. MCP Categories

4. Availability of algorithmic implementations

232 Semantic descriptions of MCPs though

KnowBaSICS-M

Web MedAl Project

Initial MCP KB

1. Know about MCPs2. Familiar with the MedAl

projec

Users Defined in MCP KB

4 Users Specialist in the fields of

cardiology and pulmonary medicine

Define Users as Knowledge Authors in MCP Ontology

1. Express their queries in a very descriptive,natural language

2. No instructions or training concerning the keywords

selection

Users instructions

Result sets of ranking MCPs

A total of 68 clinical

Questions were addressed by the

physicians

Evaluation Process

1. Users manual marks relevant MCPs residing in the MCP KB corresponded to their clinical

questions.2. Users compare the manual marking of the MCPs and the

results obtained from the system3. Create new MCPs instances -

New Knowledge

Estimation of precision and recall

futures

New MCPs in KB

Result assessment

16C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 17: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Experimental Results of Search Experimental Results of Search

18

Precision

Recall

harmonic mean

C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 18: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Test Case 1. Search similar MCP: Treatment of Test Case 1. Search similar MCP: Treatment of massive pulmonary embolism2. Find Algorithmimassive pulmonary embolism2. Find Algorithmic Sequencec Sequence

to manage a specific case to manage a specific case

19C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras

Page 19: Usage of Semantic Web Technologies (Web 3.0) Aiming to Facilitate the Utilisation of Computerized Algorithmic Medicine in Clinical Practice [Med2 Bratsas V2]

Future Research Future Research

Major technical challenge is the automated incorporation of the content located at existing repositories such as MedAl in the MCP KB (wrapper-mediation based)

An extension of KnowBaSICS-M is considered to support the automated identification of individualised algorithms that will be linked with Electronic Health Record (EHR) data (Archetype - OpenEHR),

High quality medical education (Problem Based Learning & Case Based Learning - HealthCare LOM -SCORM)

Semantic Wiki about algorithmic medicine◦ combination of web-2.0 and Semantic Web (e.g. wiki

professional)

20C Bratsas, P Bamidis*, E Kaimakamis, N Maglaveras


Recommended